Making MGAIC
generative AI research with real-world impact
Advancing the frontiers of generative AI
MIT researchers are pushing the boundaries of model architecture, safety, and alignment to enable generative AI systems that are more capable, trustworthy, and open. These breakthroughs support discovery and creativity across science, health, education, and the arts.
Designing AI that elevates human work
MGAIC supports the development of tools that amplify human intelligence—enhancing productivity, creativity, and decision-making in real-world domains. The goal is not to replace people, but to create AI that collaborates, augments, and empowers.
Engineering for scalable, responsible deployment
Scaling generative AI requires addressing critical infrastructure challenges—from compute efficiency and power consumption to data integrity and system robustness. MIT’s cross-disciplinary expertise helps design AI systems that are both powerful and sustainable.
Expanding access through education
To ensure AI benefits are broadly shared, MGAIC promotes open-source tools, new models of learning, and global collaborations. With a focus on inclusion and opportunity, we are helping shape a future where everyone can participate in AI innovation.
“The remarkable progress in generative AI we’ve seen over the past year has been fueled by advances in fundamental science and engineering — areas where MIT excels.”
— Sally Kornbluth, President of MIT
Our founding members
Drive the strategic direction of the Generative AI Impact Consortium and fund project ideas from MIT’s research community.
Student impact
The future of generative AI will be shaped by today’s students—and the MIT Generative AI Impact Consortium is committed to giving them a central role in that future.
Through research opportunities we support undergraduate and graduate students in tackling real-world challenges with generative AI. These opportunities empower students to collaborate with faculty and industry mentors, work on projects with societal significance, and gain hands-on experience at the forefront of AI research.
From developing open-source tools to exploring ethical deployment frameworks, students in the Consortium are not just learning about the future of AI—they’re building it. Their work spans disciplines, connects communities, and accelerates progress toward responsible, cross-sector solutions.
Funded projects
Voices of the poor
Team: Sendhil Mullainathan (EECS; Economics), Ashesh Rambachan (Economics), and Oeindrila Dube, University of Chicago
Area: Social Impact / Economics / AI & Society
This project explores how generative AI can help uncover, synthesize, and elevate the lived experiences of low-income communities that are often excluded from traditional policy design. By analyzing qualitative data and generating actionable insights, the team aims to build tools that policymakers and researchers can use to better understand and serve vulnerable populations.
Enabling inductive reasoning
Team: Rafael Gómez-Bombarelli (DMSE) and Ju Li (NSE), Kaiming He (EECS)
Area: Science & Engineering / Materials / AI Reasoning
This flagship project pushes the frontier of generative models by embedding principles of scientific reasoning. The team is developing new architectures that go beyond data fitting to enable AI to propose hypotheses, simulate outcomes, and guide experimental design. The goal: accelerate discoveries in physics, chemistry, and materials by teaching AI to reason like a scientist.